Re: Jabberwocky/AI Language Learning
From: | David J. Peterson <dedalvs@...> |
Date: | Monday, March 19, 2007, 17:09 |
Carsten wrote:
<<
Isn't this how children actually acquire language? They
learn possible patterns to form grammatically acceptible
sentences on the one hand, but they also get a feeling for
what 'sensible' utterances are as they get older, either due
to learning how things in their environment work or due to
their parents telling them so.
>>
That's what I think (as opposed to some notion of UG, parameter
setting, etc.), and I was pretty sure that I'd seen a language-
learning algorithm that did just this--in fact, I'm pretty sure a
grad. student at UCSD created something that did this to see how
well a machine could learn languages with varying word order
types, accusativity, and case marking. (Incidentally, it found out
that it's more difficult to learn languages where objects come
before subjects, and slightly more difficult to learn ergative
languages.) That's why I thought that maybe that's what the
Jabberwacky program was doing.
-David
*******************************************************************
"sunly eleSkarez ygralleryf ydZZixelje je ox2mejze."
"No eternal reward will forgive us now for wasting the dawn."
-Jim Morrison
http://dedalvs.free.fr/
Reply